Multispecies occupancy modeling
Sampling for individual organisms may result in imperfect detection. Occupancy models can account for imperfect detection by estimating the probability of detection as well as the probability that species occur at a location. While many occupancy models focus on single species, this project developed a multispecies occupancy model that allows for the estimation among species occupying locations across space and time.
It is sometimes important to know where species exist and the probability a species occupies a specific location. However, when sampling for individuals, some may not be observed. Occupancy models account for this imperfect detection by estimating the probability of detecting a species within a location. For example, a fish biologist may sample a lake for X, Y, and Z species. A species is considered detected if they were able to collect or observe the target organism. Lack of a positive detection may mean that the biologist missed the species using their specific sampling approach or the species was not present. Occupancy models estimate this probability. and multispecies occupancy models can account for these relations among species.
Predictive occupancy models are challenging to develop, and USGS scientists used a probabilistic programming language called Stan to conduct their own research and allow other scientists or managers to calculate multispecies locations more accurately. The software package also shows how the models were programed including tutorials explaining the code and math supporting the models.
occstanhm: Hierarchical occupancy models with correlated error structure
occStan: Occupancy models with RStan
Sampling for individual organisms may result in imperfect detection. Occupancy models can account for imperfect detection by estimating the probability of detection as well as the probability that species occur at a location. While many occupancy models focus on single species, this project developed a multispecies occupancy model that allows for the estimation among species occupying locations across space and time.
It is sometimes important to know where species exist and the probability a species occupies a specific location. However, when sampling for individuals, some may not be observed. Occupancy models account for this imperfect detection by estimating the probability of detecting a species within a location. For example, a fish biologist may sample a lake for X, Y, and Z species. A species is considered detected if they were able to collect or observe the target organism. Lack of a positive detection may mean that the biologist missed the species using their specific sampling approach or the species was not present. Occupancy models estimate this probability. and multispecies occupancy models can account for these relations among species.
Predictive occupancy models are challenging to develop, and USGS scientists used a probabilistic programming language called Stan to conduct their own research and allow other scientists or managers to calculate multispecies locations more accurately. The software package also shows how the models were programed including tutorials explaining the code and math supporting the models.